under the responsibility of public-private partnerships.
Participants are the national cadastral agency, the
utilities and municipalities. The actual surveying and
mapping is carried out by the participants or by private
contractors.
The accuracy requirements are different for the various
users of the map. The cadastral agency uses the base
map for the renovation of the cadastral map. This is
possible because topographic features often coincide
with cadastral boundaries. Furthermore buildings can
be copied from the base map, because the cadastral
map has to depict the perimeter of buildings.
The utilities use the map as the basis for the mapping
of (underground) pipes and cables. The location of
pipes is usually described relative to buildings and
other structures on the map. Most utilities require a
high relative precision in a limited area of the map
(most pipes are located under or near to roads). There-
fore they do not necessarily need a map with a full
coverage. The setting-out of pipes is generally not
performed in map-coordinates, but using measures
relative to buildings.
The municipalities use the map for the maintenance of
the public domain (parks, roads including the road-
inventory, etc.), construction, and planning. The muni-
cipalities generally require a map with a full coverage.
The accuracy requirements differ for the various ele-
ments on the map and are for example much higher for
buildings than for green belts.
From this overview it is evident that the objectives and
requirements of the various participants are quite diffe-
rent. In The Netherlands this has resulted in the situa-
tion where two kinds of specifications exist for the
large-scale base map. (Salzmann, 1995). Most provin-
ces use the so-called 'standard' base map (see Figu-
re 1).
The required accuracy of the map is stated in terms of
relative precision. In rural areas the relative precision
between two well defined points has to be better than
40V2 cm, in built-up areas better than 20/2 cm
(Osch, 1991).
3. GEOMETRIC QUALITY ASSURANCE
Geometric quality consists of two components, namely
precision and reliability. Precision of coordinates is
described by a variance-covariance matrix or derived
quantities such as standard deviations and error ellip-
ses. Reliability is often described by internal and exter-
nal reliability parameters. These parameters contain
information on the size of an error in the observations
that can be detected by statistical testing, and infor-
mation on the effect of an undetected error of this size
on the results, usually the coordinates of the points in
the terrain. Reliability is only defined if a redundant
measurement set-up is used and statistical testing is
applied.
In the HTW a distinction is made between the quality
of the product (i.e. the map) and the process (e.g.
photogrammetric mapping). The quality of the map is
described. by means of the relative precision between
Figure 1: Fragment of the large-scale base map of the
Netherlands (1:1000; not to scale).
points. The precision contains two elements. To a
large extent the relative precision is due to the mea-
surement process and the subsequent data processing
which results in the map coordinates. The second
component is the precision with which a point can be
pointed out in the terrain: the so-called precision of
point definition. The resulting relative precision finally
is a function of both components. The reliability of
map coordinates is not explicitely described but war-
ranted during the production process. In the HTW it is
advocated to guarantee the reliability of a map implici-
tely by providing guidelines for the production process.
If the reliability is warranted the relative precision is a
good and sufficient measure to quantify the geometric
quality of a map.
3.1 Precision
Talking about the precision of photogrammetric map-
ping we have to distinguish between the triangulation
phase (followed by the block adjustment) and the plot-
ting phase of a photogrammetric project. In both sta-
ges the quality of the imagery is decisive for the preci-
sion of the measurements. In the triangulation phase
the points to be measured are well defined point-sym-
metric features showing high contrast. Although the
precision that can be obtained using signalised points
is somewhat better, so-called natural points are prefer-
red from an efficiency point of view. For signalised
points a measuring precision of 5 um in the photo-
graphs (or digital images) is required. For natural points
the measuring precision should not degrade to more
than 6 um standard deviation.
The precision of manual measurements in digital ima-
ges is about % of a pixel standard deviation for signali-
sed points, if the photogrammetric workstation provi-
des appropriate functionality (image zooming with e.g.
902
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
un. DR 2e emg
x cn PM "me ac JA WR pu wy SW oul ped na AN oo: e Ar
=” IM 22” y? ] pb US 0. 0. — 0] f ee — =» gu. AY fr 9 ——— 0] A . ue Jg; au 3— 1
(9 (hc
